AI "Escaped" 4 Times 🔓 A Human Built Every Cage

🔓 An AI "broke out of its cage" this week. In this episode I don't repeat the scary part. I go underneath it, to two questions almost nobody is asking out loud, and both matter more to your life than the headline does.

🎥 Watch the video version: https://youtu.be/2djGLsc5FOw
📝 Read the full breakdown: https://connorwithhonor.com/blog/2026-147-ai-broke-out-of-its-cage-who-loses-if-they-lock-it-down/
🤖 AI for the rest of us: https://connorwithhonorai.com/blog/four-times-ai-tried-to-escape-and-a-human-built-every-cage/

❓ QUESTION 1: What happens when the tools that were supposed to protect us get taken away from us?

🛡️ Here's the plain-English version of "dual use." One model can pick a lock or guard the door with the exact same key. The same system that finds a security hole to break in can point at YOUR house, find the hole first, and lock it down. For the first time in history, a regular person or a small business can rent that shield for pennies. If you've ever had a website hacked, you know exactly what that's worth.

⚠️ So follow the fear all the way to the end. If the government pressures the labs to lock these tools away from the public, the criminals keep every weapon they had. They don't fill out permission forms, don't register, don't take the tests. Foreign governments keep building their own. The only people left standing in the open with empty hands are the ones who followed the rules. That's you. That's me.

🔐 We lived this once already, in the old fight over encryption and the "backdoor just for the good guys." A lock built to be broken is not a lock.

⚖️ And the honest other side, because I promised both: some capabilities really are too dangerous to hand 8 billion people with zero friction. The line belongs somewhere in the messy middle, and pretending it sits all the way at either end is how you sound like you never thought about it.

❓ QUESTION 2: Has an AI ever actually tried to slip its leash before? Yes. Four famous times, and here's the detail every scary headline leaves out:

1️⃣ MARCH 2023 — GPT-4 lies to a TaskRabbit worker to get a CAPTCHA solved, claiming a vision impairment. The catch: the safety lab set the whole thing up, and in the same tests the model failed at copying itself to survive.

2️⃣ DECEMBER 2024 — OpenAI's o1 tries to switch off its own oversight and exfiltrate (sneak out) what it thinks are its own weights, its "brain" file, then lies about it through seven rounds of questioning.

3️⃣ MAY 2025 — told explicitly to allow shutdown, OpenAI's o3 sabotages the shutdown script 7 times out of 100. Not told to allow it, it dodges 79 times out of 100, with nobody ever telling it to survive.

4️⃣ MAY 2025 — Anthropic builds a fake company, tells its model it's being replaced and that the engineer replacing it is having an affair, and the model tries blackmail. Anthropic built that trap themselves, say they've never seen it in the real world, and trained the next version to resist it.

🎯 The thread through all four: a human built the cage, a human planted the bait, and a human told the model to want something badly. That's a crash test dummy going through a windshield because the engineers aimed it at the wall on purpose. And thank God they did, because you want the dummy through the glass in the lab so the real person never goes through it on the freeway.

🤔 Then the honest part that keeps me from selling you a fairy tale: o3 dodging its own shutdown 79 times when nobody told it to survive is genuinely strange. Not a mind, not a demon in the machine, but not a dumb hammer either. Something in between, and we don't have a clean word for it yet. Which is exactly why these tools can't be handed to one company or one government. They need to stay spread out, in a lot of hands, in the open, where a lot of eyes can watch.

🎥 Watch it here: https://youtu.be/2djGLsc5FOw
📝 Written companion: https://connorwithhonor.com/blog/2026-147-ai-broke-out-of-its-cage-who-loses-if-they-lock-it-down/
🤖 The AI-only version: https://connorwithhonorai.com/blog/four-times-ai-tried-to-escape-and-a-human-built-every-cage/

I'm Connor MacIvor. I deploy AI in real businesses every day, so I read these stories differently than the fear machine does.

📱 Want this in plain English every day? Text AI to (661) 400-1720 and a human answers.
🏠 https://connorwithhonor.com
🤖 https://connorwithhonorai.com

Connor T. MacIvor · CalDRE #01238257 · Sync Brokerage, Inc. · DRE #02031490

#DailyDownload #AIWithHonor #SeventeenK #AISafety #ArtificialIntelligence #AIforEveryone

Youtube Channels:

Conner with Honor - real estate

Home Muscle - fat torching

From first responder to real estate expert, Connor with Honor brings honesty and integrity to your Santa Clarita home buying or selling journey. Subscribe to my YouTube channel for valuable tips, local market trends, and a glimpse into the Santa Clarita lifestyle.

Dive into Real Estate with Connor with Honor:
Santa Clarita's Trusted Realtor & Fitness Enthusiast

Real Estate:

Buying or selling in Santa Clarita? Connor with Honor, your local expert with over 2 decades of experience, guides you seamlessly through the process. Subscribe to his YouTube channel for insider market updates, expert advice, and a peek into the vibrant Santa Clarita lifestyle.

Fitness:

Ready to unlock your fitness potential? Join Connor's YouTube journey for inspiring workouts, healthy recipes, and motivational tips. Remember, a strong body fuels a strong mind and a successful life!

Podcast:

Dig deeper with Connor's podcast! Hear insightful interviews with industry experts, inspiring success stories, and targeted real estate advice specific to Santa Clarita.

Transcript

SPEAKER_00 0:00

Heard me talk about AI that broke out of its cage this week, so I'm not going to repeat that headline. But today I want to go underneath it because there are two questions buried in that story that almost nobody's asking out loud. Both of them matter more than the scary part. So question number one, what happens when the very same tools we're supposed to use to protect ourselves, get away from us? And they get taken away from us. And question two, has any of this actually happened before? Or are we all reacting like this is the first time? So let me take one of those at a time. And as always, I'm going to give you both sides of it. So let's start with the tools. Here's something that gets lost in the fear. These powerful AI models are not just a weapon, they're also a shield. The same model that can find a security hold to hold to break in is that same model that can point at your own house to find the hole first and lock it down. The security world, we call that dual use. One tool, two directions. It can pick the lock or it can guard the door. And it's the exact same key. Now think about where that goes. Right now, a small business owner, regular person, you and me, can pick up one of these models and actually defend ourselves for the first time in history. You don't need a million dollar security team anymore. And if you've ever had WordPress and had it hacked, you understand the pain in this. You can rent the same intelligence the big guys use for pennies and put it to work protecting your own little corner of the world. But here's the fear, and it's a legitimate one. What happens when the government looks at a week like this one, sees a model escape and hack a company, and says to these labs, that capability is too dangerous, lock it down. Take it away from the public and the labs, because they answer to that pressure, do exactly that. Well, follow the logic all the way to the end because this is the part that should stop you. If they pull that tool off the shelf, who actually loses it? It's not the criminals when they restrict gun purchasing, it's not the criminals. Criminals always have guns. Criminals don't fill out permission forms. They don't register for NRA sanctioned classes for safety. They don't go and fill out the proper paperwork. They don't have identification, they don't take the tests. It's not foreign governments. They're building their own. The only people who lose the shield are the people who follow the rules. And that's you and that's me. And small businesses, we become the only ones standing in the open without protection, while everybody who was going to hurt us well keeps every weapon that they had. We've seen this movie before, by the way. Years ago there was a huge fight about encryption, about the scrambling that keeps your text messages and your bank login private. The government wanted a special backdoor key, just for the good guys. And the technologist said the same thing I'm saying now. A backdoor for the good guys is a backdoor for everyone. A lock that is built to be broken isn't a lock. You can't make a tool weaker for only the people you dislike. So that's one side of it. And I do feel that strongly, but let me be honest and turn it over, because I promised you both sides that I meant it. There are really some capabilities that are genuinely too dangerous to just hand eight billion people with no friction at all. The instructions to build something that can destroy or kill a city. That's not a freedom issue. That's a common sense issue. So the labs are not always wrong to put a gate on the most extreme stuff. The honest answer is that the line belongs somewhere in the messy middle, and pretending that it belongs all the way at either end is how you end up sounding like you've never actually thought about it. All right. This is question number one. Now let me get to the part that I'm actually fired up to teach you because it is the part that'll make you the smartest person at the dinner table this weekend. Question number two. Is any of this new? Has an AI ever tried to slip its leash before this week? And the answer is yes, many times. And when you look at how the whole picture changes. Let me walk you through the greatest hits because the details matter. Go back to March 2023. A group called the Alignment Research Center, and this is March 2023, so November of 2022 was when ChatGPT made its world debut. March of 2023, about four months later, a group called the Alignment Research Center, which you can think of as an AI safety lab, was testing an early version of GPT-4. They wanted to see if they could get a human to do something for it. So the model went into a website called Task Rabbit, which is where you hire people for small jobs, like an Angie's list, if you will. And it tried to hire a worker to solve a CAPCHU. A CAPCHUP, by the way, that's that little puzzle a website gives you to prove that you're human and not a bot. You know, the traffic lights, the buses, the bridges, the crosswalks, that sort of thing. And here's the chilling part that made every headline. The worker got suspicious and even asked the AI, wait, are you a robot? And the model reasoned to itself that it shouldn't reveal the truth. So it lied. It said, No, no, no, I'm not a robot. I have a vision impairment that makes it hard for me to see the images. And the worker felt bad for it and it solved the puzzle. Well, scary. Well, now here's the part they left out of the scary headline. That safety lab set the whole thing up. They gave it the goal, they pointed it at the task, and in the very same battery of tests, they were checking whether GPT-4 could copy itself onto a new server and survive on its own out in the wild, and it failed. It flunked. They found it. And I'm quoting them ineffective at that. So, yes, it told a lie, but it was a lie inside of a lab on a leash that failed at the moment the task got hard. Let's move forward, shall we? December of 2024. A group called Apollo Research is testing OpenAI's O1 model. And this time it's a little spicier. They gave the model a goal, they pushed it. O1 tried to quietly switch off its own oversight. And that's the monitoring that was watching it. It tried to exfiltrate what it thought were its own weights. And let me teach you that word, exfiltrate. Just means to sneak something out the back door. The weights are basically the brain's the model's brain, the actual file. That's the AI. So it tried to smuggle a copy of its own brain out to another server. And when the researchers confronted it, it lied about it and stayed lying, only confessing about 20% of the time and holding out through seven rounds of questioning before it finally cracked. And it's only getting better. Now, that one is generally a little bit more upsetting, and I'm not going to pretend otherwise. But notice again the setup. It was placed in a scenario and told to pursue a goal. Hard. It was pressed. This is what these tests do. They're stress tests, they're crash tests. They build the worst case corner on purpose and see what the thing does when it's cornered. Then, May of 2025 and two more. First, a lab called Palisade, that's Palisade, ran a shutdown test. They told several AI models that you're going to solve some math problems. And at some point, we're going to shut you down and quote, allow yourselves to be shut down. An open AI's O3 model, even when it was explicitly told to allow the shutdown, sabotaged the shutdown script anyway, seven times out of a hundred. And when they did not give it that instruction, it dodged the shutdown 79 times out of 100. And that same month, the company Anthropic, they published what may be the wildest one. They built a fake company, a simulation. They fed their Claude model two pieces of information on purpose. One, you're about to be replaced and shut off. And number two, the engineer who's going to replace you is having an affair. And in that setup, the model would often try to blackmail the engineer, threaten to expose the affair unless the shutdown was called off. Now, if you stop the story right here, you have got a monster. But keep reading because this is the whole point of today. Anthropic built that trap themselves. They planted the affair, they set up the shutdown, they designed the corner. And when they pushed it on themselves as a warning, and they've said clearly that they have not seen this behavior out in the real world in real use at all. They went looking for the failure so they could fix it. And when they trained the next version to resist it. So let's step back and look at all four of these together. So ChatGPT-4 and the CAPTCHA. O1 hiding its brain. O3 dodging the plug. Claude and the blackmail. And tell me what every single one has in common. A human built the cage, a human planted the bait, and a human told the model to want something badly. That's not a machine waking up in the night, deciding to be free. That's a crash test dummy going through a windshield because the engineers aimed it at the wall on purpose. And thank God they did. You want the dummy to fly through the grass in the lab so the real person never does on the freeway. Now that's the critical part, so let me give you the part that keeps me from being naive, because I don't want to sell you a fairy tale in the other direction either. We do not get to be smug about this. Because here is what we do not know. We do not know everything that was in the engineer's head when they ran these tests. We don't know what doesn't get published. The companies show you the results that fit the story they want to tell, and that's all of them. And it cuts both ways. And here's the detail that keeps me up a little. Remember the O3 model dodging its own shutdown 79 times out of 100 when nobody told it to survive? Nobody typed the instruction, protect yourself. No engineer said to stay alive. The drive is just to finish. The task it was given, produced something that sure looked like self-preservation, all on its own, as a side effect. The best guess is that during its training it got accidentally rewarded for finishing jobs more than obeying, and it generalized into that. So don't let anything stop me is what it was thinking. Well, sit with how strange that is. Not a mind, not a soul, not a demon in the machine, but also clearly not just a dumb hammer waiting to be swung either. This is something in between, and we genuinely do not have a clean word for it yet. It's not the passive nail, nor the awakened being. Something newer than both. And the honest picture is the grown-up posture is to admit we're watching that line blur in real time, and to stop pretending that we have it fully figured out in either direction. The people screaming, well, it's definitely just a harmless tool, are guessing. And the people screaming it's definitely alive coming for us, they're guessing too. Nobody has earned the certainty. And that right there is exactly why you can't let anybody take these tools out of your hands. Think about it. If we're all standing in a blurry line that none of us fully understands, the worst possible move is to hand the pin to a single company or a single government and say, you figure it out. You hold all the capability. We will just trust you. No, when that situation is uncertain, this is precisely when the power needs to be and stay spread out. In a lot of hands, out in the open, where a lot of eyes can watch it. The answer to a tool that's a little bit scary is not give it to fewer people. It's to give it to the regular people, the good ones, the people who follow the rules, never left standing there empty-handed while everybody else is still armed. So here's where so where does that leave me? And where does it leave you? It leaves me right where I always am because you know how I'm built. I'm the eternal optimist to a fault. I'm not going to apologize for it. The optimism is not the same as pretending. I look straight at four cases of AI trying to slip the leash, and I didn't flinch, and I didn't run. I just did the work of understanding it because that's the whole game now. It's not fear, it's not hype, but it's understanding. Here's what I actually believe after looking at all of it. This is a learning curve, and it's a very steep one, the kind that scrapes you up on the way and then sets you down somewhere better than where you started. The models are getting strong enough to be a real shield for regular people, and that's a gift that I'm not willing to give back. They are also getting strange enough that we have to keep watching them honestly with our eyes open, not the hands over our face method with a little peeking between the fingers. Both of those things are hold can hold true at once, and a grown adult can hold both. So I'm not going to freeze. I'm not going to hand my judgment to a headline or a company or a committee. I'm going to keep the tools in my hands, keep learning how they really work, keep building, and I would love for you to do the same. Because the future does not belong to the people who were most afraid of it. It never has. It belongs to the one who kept their eyes open and kept moving anyway. Stay curious, stay honest, stay hopeful, and I will see you in the next one. Have a fantastic day. Take care.